Fuel moisture soft-sensor and its validation for the industrial BioGrate boiler

Jukka Kortela, Sirkka-Liisa Jämsä-Jounela Aalto University School of Chemical Technology

NPCW 2012

Contents

• Objectives • Description of the Biograte process • Fuel moisture soft-sensor • Decription of the testing environment • The test results of the fuel moisture soft-sensor • Conclusions

Objectives of the advanced control of the plant • The main objective is to control the plant and avoid

shutdowns while burning wide variety of problematic bio fuels with fluctuating fuel quality and moisture

• To keep constant power production with fluctuating fuel quality and moisture

• To react sudden changes of power demand – with fluctuating fuel quality and moisture – with long time delays

Objectives of the fuel moisture soft-sensor • The fuel moisture soft-sensor is needed to

achieve control objectives – Robust control is not efficient because fuel moisture

is not measured and process dynamics contain long delays

– Estimating fuel moisture allows control to compensate the affect of moisture variation

Description of the Biograte process • The fuel is fed onto the center of a grate from below by a

stoker screw. • Alternate rotating rings are pushed clockwise or

counterclockwise respectively.

Description of the Biograte process • asdasd

Input variable Output variable Step response Dead time Time constant Primary air flow Combustion power 30s 2s Moisture in fuel feed flow

Combustion power 15min 10min

Combustion power Drum pressure 4s 20min Combustion power Secondary

superheater 1s 1.5min

Combustion power Fuel moisture soft-sensor

4s <1min

Combustion power Combustion power Estimation

4s <1min

Description of the delays in the Biograte process • The BioGrate process is characterized by large time

constants and long dead times

Control of BioGrate boiler

Change in power demand

Hot hot water power / flow

control

Drum pressure control

Total air / Primary air

control

Primary air fan

Total air / Stoker speed

control

Stoker motor speed

Oxygen correction

control

Total air / Secondary air

control

Secondary air fan

S

PT

FT M

Drum pressure

Primary air

Stoker speed Boiler

BioGrate

Turbine

Hot water network

Contents

• Objectives • Description of the Biograte process • Fuel moisture soft-sensor • Decription of the testing environment • The test results of the fuel moisture soft-sensor • Conclusions

Fuel moisture soft-sensor • The fuel moisture disturbance w is estimated

with least squares

• The fuel moisture soft-sensor is based on following sub models

– Superheater model estimates the output enthalpy h2 of the secondary superheater

– Flue gas model estimates flue gas flow mfg and flue gas temperature Tfg

– Combustion model estimates fuel flow mf(w)

where h2 is the output enthalpy of the secondary superheater

( )[ ]sMJhmhmQVdt

hdt /ˆ1ˆ

22112 −+=

ρ

( ) ( )[ ]sMJTTkTTmQ wf gwwf gf gww /ˆˆˆ 4465.0 −+−=α

( )[ ]sMJTTmQ wct /8.02 −=α

( )[ ]sKQQCmdt

Ttw

pt

w /1

−=

Tfg ^

mfg ^

m1

Qw

h1 h2

Qt

( ) ( ) ( )∑=

−=N

i

wihihwJ0

2

22ˆmin

Fuel moisture soft-sensor: Flue gas • Flue gas flow is:

and Nfg is flue gas flow for one kilogram of fuel (mol/kg).

• Flue gas temperature is: where qf is is effective heat value of wet fuel, Fair is total air flow (m3/s), Ci is specific heat capacity (J/molT), and NExAir is excess air (mol/kg)

[ ]skgN fg /⋅mfg=mf(w)

^ ^

( )( )( ) ( )( )( )( ) ( )

2222222222

22

79.021.076.3

ˆ1041.22/79.0ˆ1041.22/21.0 33

NExAirOExAirNNg

OOHHOHSOSCOC

NfAirOfAirf

CNCNCnNCnnCnCn

CwmFCwmFq

⋅+⋅++⋅++++

⋅⋅+⋅⋅+=

−−

Tfg ^

Fuel moisture soft-sensor: Combustion • The fuel flow mf(w) is given as follows:

where XO2 is oxygen content of flue gas (%), nAir is total air flow (mol/s), NO2

g oxygen needed to burn 1 kilogram of fuel (mol/kg), and Nfg is flue gas flow for one kilogram of fuel (mol/kg)

( )[ ]skg

NNX

N

nX

gOfg

OgO

AirO

/76.4

100

10021.0

2

2

2

2

⋅−+

−

=mf(w) ^

Contents

• Objectives • Description of the Biograte process • Fuel moisture soft-sensor • Decription of the testing environment • The test results of the fuel moisture soft-sensor • Conclusions

Decription of the testing environment

• The experiments were conducted at the BioPower 5 CHP plant that produce 13.5 MW heat and 2.9 MW electricity. Two fuels were used:

1. spruce bark with the moisture content of 57% 2. dry woodchips (spruce) with the moisture content of 20%

• Dry biomass was fed manually into the screw conveyor between

wet fuel through the extra feeding box.

Decription of the testing environment

The accuracy of the fuel moisture soft-sensor was investigated by • Sampling fuel feed every 5 minutes. • Servomex 2500 FT-IR flue gas moisture analyzer. Samples were

recorded every second.

Measurement set-up for fuel feed samping

Measurement set-up for the FT-IR analyzer in the flue gas duct

Contents

• Objectives • Description of the Biograte process • Fuel moisture soft-sensor • Decription of the testing environment • The test results of the fuel moisture soft-sensor • Conclusions

The test results of the fuel moisture soft-sensor: Accuracy • According to the fuel sampling, the wet fuel contained 54.4%

moisture per kg fuel on average The fuel moisture soft-sensor content resulted in an average fuel moisture content of 54.6% per kg fuel.

• Dry fuel contained 25.9% moisture per kg fuel on average The fuel moisture soft-sensor content resulted in an average fuel moisture content of 27% per kg fuel

Fuel moisture soft-sensor (thick line), sampled fuel moisture (stars), and fuel moisture calculated from the FT-IR measurement (thin line) for comparison.

The test results of the fuel moisture soft-sensor: Dynamic behaviour • The fuel moisture soft-sensor responds to step changes within 1 minute

compared with FT-IR according to the test results and cross-correlation analysis

Fuel moisture soft-sensor (thick line), sampled fuel moisture (stars), and fuel moisture calculated from the FT-IR measurement (thin line) for comparison.

Grate temperatures during the test. The grate rings are numbered from the center (Grate ring 2) to the edge of the grate (Grate ring 12)

Conclusions

• Fuel moisture soft-sensor opens new possibilities to control combustion: – Fuel moisture soft-sensor shows a decrease in fuel moisture

20 minutes before drum pressure drops – Fuel moisture soft-sensor responds to step changes

within 1 minute compared with FT-IR.

• Fuel moisture soft-sensor value and the sampled moisture content matches well

• In 15 MW BioPower CHP plant average 0.5 million euros is saved yearly